An optimal packing problem of non-convex polygons on a sheet is considered and applied to a textile nesting problem. We propose a procedure of grouping objects and adaptive meta-heuristics. The grouping procedure classifies objects into groups which are then allocated on the sheet using adaptive meta-heuristics. As a result, effectiveness of the meta-heuristics is improved. The performance of the proposed method is compared with other meta-heuristics using simulation experiments. The result shows that the present method is superior to cases where adaptive meta-heuristics without grouping are used.
IntroductionWith the remarkable increase of computer capability, new research in optimization algorithms of meta-heuristics is becoming more popular and much research has been done on real applications. Three major methodologies in metaheuristics are simulated annealing [1], genetic algorithms [2], and tabu search [3,4]. They should be remarked upon for their practical usefulness as well as fundamental ideas. Unlike the traditional methods of optimization on the basis of rigorous mathematics, they use heuristics whereby obtained solutions are, in general, not strictly optimal. However, the methodologies are simple, flexible, applicable to many real complex problems, and produce practical solutions.A notable feature of these methodologies is that they are based on different fundamental paradigms. Simulated annealing is based on a physical paradigm while genetic algorithms are based upon a biological paradigm. We can remark that a human paradigm exists under the tabu search, since human logic such as taboo is used, and moreover strategic search is used in contrast to a blind search in simulated annealing and genetic algorithms.We enhance the human features in the algorithms proposed in this paper. The proposed algorithms have adaptation mechanisms whereby search effectiveness is improved. These adaptation mechanisms use human logic and memories. By using such human mechanisms, the whole system becomes more human-oriented. We call these mechanisms adaptive meta-heuristics which have a number of options: adaptive local search, adaptive simulated annealing, and adaptive tabu search.The aim of the present paper is not to propose a general methodology. Rather, our purpose is to provide practical solutions for the real-world problem of the optimal allocation of irregular objects. The objects are allocated on a sheet such that waste is minimized. Such a problem arises in many industrial fields, such as the textile industry and the sheet metal industry [5,6]. This so-called nesting problem is NP-complete [7], hence no theoretically efficient algorithm seems to exist. However, there have been several approaches, such as linear programming [8], expert system [9], heuristic procedure [10], and metaheuristic approach [11,12]. The meta-heuristic approach has shown good performance, but in order to tackle problems of complicated shapes, further improvement to the algorithm is required.The present method consists of two stages. In the f...